2.1 CaeB localizes at the ER

To determine the subcellular localisation and molecular activity of CaeB, we chose the plant Nicotiana benthamiana as a model system. We hypothesised that CaeB is also functional in planta, since CaeB acts downstream of Bax (Klingenbeck et al., 2013) and Bax induces cell death in planta (Lacomme & Santa Cruz, 1999). In addition, the plant model system allowed us to concentrate on evolutionary conserved pathways. To investigate the subcellular localisation of CaeB in planta, N. benthamiana leaves were infiltrated with agrobacteria containing plasmids to express GFP-tagged CaeB or an established ER marker fused to mCherry (Nelson, Cai, & Nebenfuhr, 2007). Two days post infiltration leaf discs were analysed by confocal laser scanning microscopy. Co-expression of both proteins resulted in a clear co-localisation at the ER (Figure 1a) similar to the results by Rodriguez-Escudero et al. (2016). As ER-mitochondria contact sites are involved in cell death modulation (van Vliet, Verfaillie, & Agostinis, 2014) and co-localisation of CaeB with mitochondria has been shown (Carey et al., 2011), we infiltrated N. benthamiana leaves with agrobacteria containing plasmids to express GFP-tagged CaeB and the established mitochondria marker IVD (Isovaleryl-coenzyme A dehydrogenase) fused to mCherry. Confocal microscopy analysis showed neither co-localisation nor association of GFP-CaeB with IVD-mCherry (Figure 1a). Similarly, in HeLa cells, we could detect partial co-localisation of GFP-CaeB with the ER marker calnexin, but not with mitotracker (Figure S1). The Pearson's correlation coefficient was 0.6 for GFP-CaeB and calnexin, but 0.0 for GFP-CaeB and mitotracker (Figure S1). These data suggest that GFP-CaeB partially localizes to the ER.

Next, we analysed the subcellular localisation of CaeB when translocated by C. burnetii. Therefore, we generated a C. burnetii strain expressing 3xFlag-tagged CaeB (Figure 1b) and determined its localisation in infected mouse embryonic fibroblasts (MEFs) by confocal microscopy. A signal for 3xFlag-CaeB was hardly detectable in the infected cells. However, if we detected a signal it showed ER localisation (Figure 1c), demonstrated by co-localisation with the ER marker calnexin. Thus, translocated CaeB also localizes to the ER during infection.

2.2 CaeB suppresses cell death in N. benthamiana induced by Bax or Pto + AvrPto

To determined whether CaeB also suppresses cell death in planta, we analysed the effect of CaeB expression in N. benthamiana challenged with different cell death inducing proteins. First, we investigated if CaeB interferes with Bax-induced cell death in plants, as in mammalian cells CaeB inhibits staurosporine-induced apoptosis downstream of Bax-activation (Klingenbeck et al., 2013). N. benthamiana leaves were transiently transformed using agrobacteria to mediate the expression of Bax in the presence or absence of GFP-tagged CaeB. Since the presence of Bax results in a rapid cell death response, Bax expression was kept under control of a dexamethasone (Dex) inducible promotor, which was activated 1 day post infiltration by spraying leaves with Dex. As expected, leaves expressing Bax showed a strong and fast induction of cell death, which was visible by the formation of necrotic lesions (Figure 2a). In contrast, co-expression of GFP-CaeB resulted in a clear suppression of Bax-induced cell death as there were no cell death symptoms visible after 2 days post Bax induction (Figure 2a). To quantify the degree of cell death over time, leaf discs were sampled and carefully transferred to water. Subsequently, the loss of membrane integrity of leaf discs was measured as changes in conductivity of the surrounding water, which is indicative of ion leakage. This confirmed rapid induction of Bax-induced cell death and its suppression by CaeB as indicated by the significantly lower increase in conductivity (Figure 2b). Thus, CaeB suppresses Bax-induced cell death in mammalian cells (Klingenbeck et al., 2013) and in planta (Figure 2a,b).

Next, cell death was induced by co-expression of the Pseudomonas syringae type III secretion system (T3SS) effector protein AvrPto and the tomato serine/threonine kinase Pto which recognises AvrPto; leading to the induction of R-gene mediated cell death in N. benthamiana (Frederick, Thilmony, Sessa, & Martin, 1998). However, how the AvrPto-Pto-interaction leads to R-protein activation, and how this causes cell death remains still elusive (Balint-Kurti, 2019; Martin, 1999). It is known that the kinase activation domain of Pto is important for AvrPto recognition (Frederick et al., 1998). In addition, kinase activity of Pto is required for the induction of the HR response (Sessa, D'Ascenzo, & Martin, 2000), for example, by Pto-mediated phosphorylation of Pti1 (Zhou, Loh, Bressan, & Martin, 1995).

In fact, co-expression of both proteins caused a clear induction of cell death, which was inhibited by the presence of GFP-CaeB (Figure 2c). Visual changes were confirmed via ion leakage measurement of leaf discs expressing Pto and AvrPto with or without GFP-CaeB (Figure 2d). These results demonstrate that expression of CaeB suppresses cell death in N. benthamiana induced by different stimuli, indicating that CaeB might target an evolutionarily conserved pathway to inhibit cell death. Although, it is currently unknown, which cell death pathways are induced by Bax or AvrPto-Pto expression in N. benthamiana.

2.3 CaeB inhibits ER stress-induced cell death in N. benthamiana

Since CaeB co-localizes with the ER and has global cell death suppression activity we analysed if CaeB interfered with ER stress-induced cell death in N. benthamiana. Leaves expressing the ER-mCherry marker alone or together with GFP-CaeB were treated with tunicamycin (TM). After 24 hr, structural changes in the ER network were inspected with a confocal laser scanning microscope. TM treatment led to the disruption of the typical ER network structure in ER-mCherry expressing leaves, whereas only minor changes in the net-like ER structure were detectable in cells expressing GFP-CaeB (Figure 3a). To clarify that this is not only a general consequence of the CaeB-mediated inhibition of cell death, we analysed the ER structure in TM-treated N. benthamiana leaves expressing the Pseudomonas syringae T3SS effector protein HopG1. Like CaeB, HopG1 was shown to prevent Bax-induced cell death in N. benthamiana (Block et al., 2010; Jamir et al., 2004). However, HopG1 did not prevent disruption of the ER network after TM treatment (Figure S2). Thus, CaeB prevents ER stress and, thus, directly protects the cell from TM-induced structural disturbances in planta.

We further investigated the role of CaeB in ER stress signalling in planta by introducing prolonged ER stress. N. benthamiana leaf discs expressing either GFP or GFP-CaeB were floated on an aqueous solution containing either TM or DMSO as a control and cell death development via ion leakage measurement was monitored over a period of 72 hr. Under control conditions leaf discs expressing GFP or GFP-CaeB showed only a slight increase in conductivity over time. In contrast, GFP expressing leaf discs treated with TM displayed a stronger increase in conductivity (Figure 3b) compared to GFP-CaeB expressing leaf discs, indicating that CaeB reduces ER stress-induced cell death in planta (Figure 3b).

2.4 CaeB suppresses cell death in mammalian cells

Intrigued by the results obtained in N. benthamiana, we analysed the ability of CaeB to suppress ER stress-induced cell death in mammalian cells. Therefore, HEK293T cells stably expressing GFP or GFP-CaeB (Figure S3) were treated with TM and the cleavage of poly (ADP-ribose) polymerase (PARP) was analysed as a well-known readout for apoptosis induction (Oliver et al., 1998). TM treatment led to PARP cleavage in GFP expressing cells, but not in GFP-CaeB cells (Figure 4a,b), indicating that CaeB prevents ER stress-induced apoptosis. In addition, the expression of GFP-CaeB prevents PARP cleavage after staurosporine (Klingenbeck et al., 2013) or etoposide treatment (Figure 4c). Both agents activate apoptosis via the mitochondrial apoptotic pathway, suggesting that CaeB might influence not only ER, but also mitochondrial activity.

2.5 CaeB modulates IRE1 signalling during ER stress conditions in HEK293T cells

In mammalian cells, ER stress triggers the activation of the unfolded protein response (UPR), which is controlled by three major sensors: IRE1, PERK and ATF6 (Walter & Ron, 2011). Therefore, we analysed the effect of CaeB on PERK, ATF6 and IRE1 signalling. Stable GFP or GFP-CaeB HEK293T cells were treated with TM for different periods of time and western blot analysis using antibodies against ER stress marker proteins were used to determine the role of CaeB in ER stress responses. First, we analysed whether CaeB alters PERK signalling. Activation of PERK leads to oligomerisation and trans-autophosphorylation (Hetz & Papa, 2018). Active PERK phosphorylates eIF2a on serine 51, which results in global repression of translation initiation (Walter & Ron, 2011). However, some transcripts (e.g., ATF4) are translated more efficiently during PERK-dependent translation repression. ATF4 is normally inefficiently translated, whereas translation is very efficient in response to PERK activation (Hetz & Papa, 2018). Thus, we analysed the level of ATF4 in cells expressing either GFP or GFP-CaeB after ER stress induction. As shown in Figure 5a, TM treatment led to a similar accumulation of ATF4 in the presence of both GFP and GFP-CaeB, indicating that CaeB does not influence ATF4 protein levels.

Next, we determined whether CaeB alters ATF6 signalling during ER stress. The transmembrane glycoprotein ATF6 is exported to the Golgi upon ER stress and is cleaved by two Golgi-resident proteases, releasing its cytosolic domain (Haze, Yoshida, Yanagi, Yura, & Mori, 1999). Under ER stress, the cleaved cytosolic domain of ATF6 translocates to the nucleus to induce UPR gene expression (Yamamoto et al., 2007). As shown in Figure 5b TM treatment led to similar cleavage of ATF6 in cells expressing GFP or GFP-CaeB, indicating that CaeB does not influence ATF6 activation.

We also analysed the influence of CaeB on IRE1 signalling during ER stress. TM treatment of GFP and GFP-CaeB cells led to splicing of XBP1. However, splicing of XBP1 mRNA was significantly enhanced in GFP-CaeB expressing cells at earlier time points after TM induction compared to GFP control cells (Figure 5c,d). In addition, the amount of XBP1s protein was increased in GFP-CaeB cells following TM treatment as compared to GFP expressing cells (Figure 5e). Thus, GFP-CaeB cells showed an earlier and higher rate of XBP1 mRNA splicing compared to GFP expressing cells after ER stress induction in mammalian cells. As activation of IRE1 may also lead to phosphorylation of downstream targets like p38 we analysed activation of p38 after TM treatment. The expression of GFP-CaeB resulted in a reduced p38 phosphorylation after TM treatment compared to controls (Figure 5f). Taken together, these data demonstrate that CaeB exclusively modulates the IRE1 signalling branch during ER stress while leaving the signalling of the UPR arms PERK and ATF6 unaltered.

2.6 CaeB modulates IRE1 signalling during ER stress conditions in N. benthamiana

UPR signalling is mediated by different ER stress sensors of which IRE1 was shown to be the most conserved between species (Chen & Brandizzi, 2013). Because we showed that CaeB expression modulates IRE1 signalling in mammalian cells and that CaeB is able to suppress ER stress-induced cell death in planta, we asked whether CaeB also modulates the activity of IRE1 in planta. IRE1 is responsible for the unconventional splicing of bZIP60 mRNA in N. benthamiana (Nagashima et al., 2011). Thus, the splicing of bZIP60 mRNA, the functional homologue of the mammalian XBP1, was analysed after TM treatment of N. benthamiana leaves expressing either GFP or GFP-CaeB. GFP-CaeB expressing leaves showed a significantly higher amount of spliced bZIP60 mRNA at early time points after ER stress induction (Figure 6a), indicating that CaeB alters the RNase activity of IRE1.

Next, we asked whether the CaeB-mediated increase in the RNase activity of IRE1 is required for the inhibition of ER stress-induced cell death. To specifically inhibit RNase activity of IRE1 without interfering with dimerization and phosphorylation of IRE1 we used 4μ8C, a small molecule inhibitor (Cross et al., 2012), which was successfully used before in Arabidopsis thaliana seedling (Zhang, Zhang, & Wang, 2016). As 4μ8C has never been used in N. benthamiana we initially adjusted experimental conditions in which 4μ8C prevents bZIP60 mRNA splicing without influencing cell viability. As shown in Figure S4 concentration above 2 μM and below 5 μM fulfils this requirement. Therefore, N. benthamiana leaf discs expressing either GFP or GFP-CaeB were floated on an aqueous solution containing TM or DMSO as a control with or without 2.5 μM 4μ8C and cell death development was analysed after 20 and 80 hr of treatment via conductivity measurement. As shown in Figure 6b, GFP-CaeB expressing leaf discs showed a significantly lower increase in conductivity after TM treatment as compared to GFP expressing leaf discs, confirming the results shown in Figure 3b. Simultaneous treatment of TM and 4μ8C strongly increased the conductivity in GFP-CaeB producing leaves, indicating that the RNase activity of IRE1 is important for CaeB-mediated inhibition of ER stress-induced cell death.

2.7 C. burnetii infection alters IRE1 signalling during ER stress

Our results indicate that CaeB modulates ER stress signalling by altering the signalling of the conserved ER stress sensor IRE1. Therefore, we wanted to address the question if IRE1 signalling is also modulated during C. burnetii infection. Therefore, HeLa229 cells were infected with C. burnetii and 3 days post-infection ER stress was induced by TM treatment. Importantly, C. burnetii infection alone did not induce activation of p38 (Figure 7a) or splicing of XBP1 mRNA (Figure 7b). However, treatment with TM resulted in a reduced activation of p38 and increased splicing of XBP1 mRNA in C. burnetii infected cells compared to uninfected cells (Figure 7a,b). This led us to conclude that C. burnetii infection modulates IRE1 activity similar to CaeB in TM treated cells.

2.8 C. burnetii-mediated modulation of ER stress signalling partially depends on CaeB

To investigate the role of CaeB in C. burnetii-mediated modulation of IRE1 signalling during ER stress, a CaeB deletion mutant (ΔcaeB) was generated using a newly developed nutritional selection system, which is based on lysine auxotrophy (Beare, Jeffrey, Long, Martens, & Heinzen, 2018). The ΔcaeB mutant is capable of growing in axenic culture, infecting and forming replicative CCVs in HeLa cells (Figure S5a–c). We then determined whether the infection with wild type or ΔcaeB C. burnetii leads to an altered signalling of ER stress sensors in presence or absence of ER stress induction. As shown in Figure 8, infection of HeLa cells with wild type or ΔcaeB bacteria did not alter the signalling of PERK, IRE1 or ATF6 in control conditions. In contrast, induction of ER stress by TM application resulted in activation of all three ER stress sensors in infected and uninfected cells. However, infection resulted in reduced ATF4 production under ER stress, suggesting decreased activation of PERK signalling during C. burnetii infection (Figure 8a and Figure S5). Processing of ATF6 was not changed by infection after TM treatment (Figure 8b). Phosphorylation of p38 was decreased (Figure 8c and Figure S5e) and XBP1 splicing increased in cells infected with C. burnetii (Figure 8d,e), demonstrating a shift of IRE1 activity towards the pro-survival signalling arm. Infection with C. burnetii ΔcaeB also resulted in reduced ATF4 induction, suggesting that other C. burnetii factors are involved in modulation of PERK signalling. Importantly, increased splicing of XBP1 was not detectable in cells infected with the CaeB deletion mutant, demonstrating that CaeB is required for upregulation of XBP1 splicing during ER stress. Furthermore, C. burnetii ΔcaeB only partially reduced activation of p38 (Figure 8c and Figure S5e), suggesting that in addition to CaeB other bacterial proteins influence p38 activation.

Together these data suggest that CaeB is essential for C. burnetii-mediated upregulation of XBP1 splicing during ER stress. Furthermore, the data reveal that CaeB in cooperation with other factor(s) is involved in the inhibition of p38 phosphorylation. In addition, the results suggest that C. burnetii modulates PERK, but not ATF6 activation independent of CaeB during ER stress.

2.9 The ΔcaeB mutant has impaired anti-apoptotic activity

As the ΔcaeB mutant is impaired in the upregulation of IRE1 during ER stress conditions, we wondered whether the ΔcaeB mutant is influenced in its ability to prevent apoptosis induction. Thus, we infected HeLa cells with either wild type, a CaeB deletion mutant or a CaeB deletion mutant complemented with HA-tagged CaeB for 3 days. The cells were treated with TM to induce ER stress-mediated cell death or with staurosporine to induce intrinsic apoptosis. Wild type C. burnetii inhibited TM- as well as staurosporine-induced cell death (Figure 8f). In contrast, the ΔcaeB mutant only partially diminished apoptosis induced by staurosporine and completely failed to prevent TM-induced cell death. Importantly, the complemented strain restored the anti-apoptotic activity of C. burnetii (Figure 8f). This data demonstrate that CaeB participates in the anti-apoptotic activity of C. burnetii.

2.10 The ΔcaeB mutant is attenuated in the Galleria mellonella in vivo model

To unravel whether CaeB is important for C. burnetii pathogenesis, we investigated the role of CaeB using the in vivo G. mellonella model, which was used before to study C. burnetii pathogenesis (Kohler et al., 2016; Martinez et al., 2016; Norville et al., 2014). Larvae were injected in the upper right proleg with either wild type C. burnetii or the ΔcaeB mutant and survival rate was monitored over 10 days. Larvae infected with the ΔcaeB mutant survived significantly longer as larvae infected with the wild type bacteria (Figure 9a). This attenuation in virulence was not mediated by a reduced infection rate (Figure 9b). However, we observed reduced replication and reduced size of the CCV in larvae infected with the ΔcaeB mutant (Figure 9c–e), suggesting virulence of C. burnetii lacking CaeB is attenuated in vivo. Whether this attenuation is mediated by reduced IRE1 activation is unclear and needs further study.

4.1 Plant material and growth conditions

Nicotiana benthamiana plants were grown with 16 hr supplemental light and 8 hr dark cycle under greenhouse conditions with 25°C during light period and 22°C during the dark period. Light intensity was approximately 250 μmol * m⁻² * s and relative humidity of 50–60%.

4.2 Cell culture

HEK293T cells stably expressing GFP and GFP-CaeB (Klingenbeck et al., 2013) were cultured at 37°C and 5% CO₂ in DMEM (Life Technologies) with 10% heat inactivated fetal bovine serum (Biochrom), 1% penicillin–streptomycin (Invitrogen) and 1.5 mg/ml G418 (Roth). HeLa229, HeLa cells (obtained from ATCC) and mouse embryonic fibroblasts (MEFS) (Schulze-Luehrmann et al., 2016) were cultured in DMEM containing 5% FCS at 37°C and 5% CO₂.

4.3 Bacterial culture and infection

Coxiella burnetii NMII strain (RSA439 clone 4) were cultured in ACCM-2, C. burnetii ΔcaeB in ACCM-D without lysine and C. burnetii ΔcaeB::3xFlag-CaeB in ACCM-D with citrulline, but without lysine and arginine at 37°C and 5% CO₂, 2.5% O₂. For infection, HeLa299 or HeLa cells were seeded in 6-well plates at a density of 2x10⁵ and infected at a MOI of 100. One day post infection 1/10 of the cells in a 6-well were transferred in a 12-well plate and after 2 additional days ER stress was induced by TM treatment. MEF's were seeded on 10 mm coverslips in 24-well plates 1 day before infection at a density of 4x10⁴ cells/well and infected with C. burnetii NM II strain (RSA439 clone 4) or C. burnetii ΔcaeB::3xFlag CaeB at a MOI of 200 for 40 hr.

4.4 Plasmid and strain construction

GFP-fusion constructs GFP-CaeB (caeB:pK7WGF2.0) and GFP-HopG1 (hopG1:pK7WGF2.0) were generated using the gateway cloning system and a CaeB specific primer set (5′-CACCGTAACAATGGCAGGCATAGCTGCAAC; 3′-CTTATTAAATTCGGGTATGTATG) or a hopG1 specific primer set (5′-CACCGTAACAATGCAAATAAAGAACAGTCATCT; 3′-GCCGTTGTAAAACTGCTTAGA) (Karimi, Inze, & Depicker, 2002). The IVD-mCherry marker (ivd:pRB-35S-GW-mcherry) was also generated using the gateway cloning system and a specific primer set for the IVD targeting sequence (5′-CACCGTAACAATGCAGAGGTTTTTCTCCGCC; 3′-ATCTTGCGCAAACTTGGATAC). Arabidopsis pto and avrPto from Pst were cloned in the pRB-35S-HA-Strep vector using pto (5′-GCACTAGTAACAATGGGAAGCAAGTATTCTAAG and 3′-GAATTCAATAACAGACTCTTGGAGACG) and avrPto (5′-ACTAGTAACAATGGGAAATATATGTGTCGGC and 3′-GAATTCTTGCCAGTTACGGTACGGGC) specific primer sets. For GFP expression the pRB-35S-GFP vector was used (Kraner, Muller, & Sonnewald, 2017). For construction of pJC-CAT::caeB-5′3′-lysCA the 5′ and 3′ regions of caeB were amplified by PCR from NMI genomic DNA using the specific primer sets (5′-CGGTACCCGGGGATCCGAAGCTGCACTTCCATGAG and 3′-CACCCATATGCGACGCGAGCGTCGAGAGCCCAAGCCGCCGTATC) and (5′-CGTCGCATATGGGTGCGCATGTACGTCATAAGTAAAGTAGCCCGCTGCGC and 3′-GAACCTGTTTGTCGACACGCCGGACAATTCTATC), respectively. The 5′ and 3′ PCR products were cloned into BamHI/SalI-digested pJC-CAT by In-Fusion, resulting in formation of an internal NdeI site between the 5′ and 3′ fragments and creation of pJC-CAT::caeB-5′3′. The 1169^P-lysCA cassette was amplified from pJC-CAT::1169^P-lysCA (Beare et al., 2018) by PCR with specific primers (5′-GCTCGCGTCGCATATGGAGCTCGGTACCCGGGGATCC and 3′-CATGCGCACCCATATGGATTAATTAGAGAACCTGTTTGTCGAC) and cloned by In-Fusion into NdeI-digested pJC-CAT::caeB-5′3′ to create pJC-CAT::caeB-5′3′-lysCA.

For transient, IPTG inducible expression of HA-tagged CaeB in C. burnetii ΔcaeB HA-tagged CaeB was introduced between the EcoRI and KpnI site into the vector pKM244mod (Schäfer et al., 2017). Thus, a seamless cloning strategy with the template pCMV-HA-CaeB and the specific primer pair pKM_ov_HA_fwd 5′-CAGGAAACAGAATTCATGTACCCATACGATGTTCCAGA and CaeB_pKM_ov_rev 3′-GCATCTAGAGGTACCTTACTTATTAAATTCGGGTATGTA was used according to the manufactures (Thermo Fisher) protocol, resulting in pKM 244mod-HA-CaeB.

4.5 Generation of C. burnetii ΔcaeB mutant and the complemented strain

Coxiella burnetii Nine Mile phase II was electroporated with 10 μg pJC-CAT::caeB-5′3′-lysCA as previously described (Beare & Heinzen, 2014). Co-integrants were selected by culturing the bacteria in ACCM-D media lacking lysine, but containing 2% sucrose for 4 days as previously described (Beare et al., 2018). Surviving transformants were expanded in ACCM-D media lacking lysine for 7 days. After spreading the diluted culture on 0.25% ACCM-D agarose without lysine clonal ΔcaeB mutants were picked after 7 days of culture. The picked clones were expanded in ACCM-D media without lysine. Clones were analysed via PCR for deletion of caeB with the specific primer sets (5′-CAGCCAACCAATGCCGGAG; 3′-CCTGGCGCCTTGACTCTAC and 5′-TTGGCAGGCATAGCTGCAACTTGT 3′-GCGGCAAAAATGTTGCCATATCAAATAATT), and for insertion of lysCA (5′-GCTAGATTCAGGTGACTGCTC; 3′-TCAGCCCGTTATAGTCTGTTCTGC). As control the IS1111 element was used (5′-AATTTCATCGTTCCCGGCAG; 3′-GCCGCGTTTACTAATCCCCA).

Coxiella burnetii ΔcaeB were electroporated with 10 μg pKM244mod-HA-CaeB as previously described (Beare & Heinzen, 2014). Bacteria were cultivated for 5 days in ACCM-D media lacking lysine, but containing 3 μg/ml chloramphenicol. The bacteria were diluted 1:500 in fresh ACCM-D media lacking lysine, but containing 3 μg/ml chloramphenicol. After 5 days of growth, the bacteria were diluted 1:100 and after 4 days of expansion, the bacteria were pelleted for 30 min at 4500×g and 4°C. C. burnetii ΔcaeB::HA-CaeB (Figure S5f) was resuspended in DMEM, 10% FCS and 5% DMSO, aliquoted and frozen at −70°C.

4.6 Agrobacteria-mediated transient expression

For transient protein expression in N. benthamiana, A. tumefaciens strain GV2260 was used. Agrobacteria were transformed with plasmids for the expression of GFP, GFP-CaeB, GFP-HopG1, ER:mCherry (Nelson et al., 2007), Pto-HA and AvrPto-HA. Agrobacteria GV2260 containing the dexamethasone inducible pTA7002:bax vector were provided by the group of H. Uchimiya from the University of Tokyo (Kawai-Yamada, Jin, Yoshinaga, Hirata, & Uchimiya, 2001). Approximately 8 weeks old fully expanded N. benthamiana leaves were inoculated with agrobacteria using a needleless syringe. For all experiments agrobacteria were mixed at equal ratios with agrobacteria containing the silencing suppressor p19 (Silhavy et al., 2002).

4.7 Confocal laser scanning microscopy

For analysis of protein localisation and ER network disruption, N. benthamiana leaves were infiltrated with the appropriate agrobacteria and imaged under a confocal laser scanning microscope (Leica SP5 II) 2 days after infiltration.

4.8 Plant cell death assays

Nicotiana benthamiana leaves were syringe-infiltrated using A. tumefaciens-mediated transient expression with a final density of 0.4 OD₆₀₀. For co-expression of Bax and GFP-CaeB A. tumefaciens cultures were mixed at equal ratios and the expression of Bax was induced 24 hr post infiltration by spraying leaves with a 25 μM Dexamethasone solution. Cell death was analysed starting 1 day after Bax induction by either ion leakage measurement or visual symptom inspection. For co-expression of Pto/AvrPto and GFP-CaeB A. tumefaciens cultures were mixed at equal ratios. Cell death was analysed starting 4 days post infiltration by either ion leakage measurement or phenotype analyses.

4.9 Measurement of electrolyte leakage

Ion leakage was measured using a conductivity metre (Horiba B-771). N. benthamiana leaf discs of infiltrated tissue were taken with a cork borer no. 5 and floated on an aqueous solution. To monitor Bax-induced cell death; leaf discs were taken 1 day post Bax induction. Ion leakage was measured for a period of 48 hr at different time points. To obtain the absolute value of ions per leaf disc, leaf discs were incubated for 20 min at 99°C and conductivity was measured subsequently. For the Pto/AvrPto assay, leaf discs were taken 4 days post infiltration and ion leakage was monitored as described for Bax, for 4 days. For TM floating assays leaf discs expressing GFP or GFP-CaeB were sampled 1 day post infiltration and floated on an aqueous solution containing either 10 μg/ml tunicamycin (Sigma) or 0.1% DMSO as control or additionally 2.5 μM 4μ8C (Sigma) to inhibit IRE1 RNase activity. Increase in conductivity was measured for the indicated periods of time followed by determination of total conductivity. The increase in conductivity is presented either as percentage of the total conductivity (relative ion leakage) or as measured in μS/cm.

4.10 Apoptosis analysis of C. burnetii infected cells

HeLa cells on coverslips were infected with wild type, ΔcaeB and ΔcaeB::HA-CaeB C. burnetii at a MOI of 200 for 72 hr. At 24 hr post-infection cells were washed three times with PBS and supplemented with fresh medium. At 48 hr post-infection Isopropyl β D 1 thiogalactopyranoside (IPTG) was added to the media to induce expression of HA-CaeB in the complemented strain. The cells were either left untreated or were treated with 0.1 μM staurosporine (Cell Signalling) for 4 hr or with 5 μg/ml tunicamycin (Sigma) for 9 hr. The cells were fixed with 4% paraformaldehyde, permeabilised with ice-cold methanol and stained with antibodies against C. burnetii and LAMP-2. The cells were mounted using ProLong Diamond with DAPI (Thermo Fisher) to determine the percentage of infected cell with fragmented nuclei as described previously (Klingenbeck et al., 2013).

4.11 ER stress induction in mammalian cell lines

ER stress in stable GFP and GFP-CaeB HEK293T cells or HeLa229 and HeLa cells was induced by adding TM to the culture media at indicated concentrations and for indicated periods of time. HEK293T cells were seeded in 12-well plates with 1 × 10⁵ cells per well. After 40 hr incubation TM was added.

4.12 XBP1 splicing analysis

mRNA of stable HEK293T or infected HeLa229 or HeLa cells was harvested and isolated using the RNeasy Plus Kit (Qiagen), cDNA synthesis was performed using SuperScript II (Invitrogen). XBP1 was amplified using dream tag polymerase (Thermo Fischer Scientific) and XBP1 specific primers (5′-CCTTGTAGTTGAGAACCAGGAG and 3′-CCATGGGGAGATGTTCTGGAG). The amplified products were analysed on a 3% agarose gel. Splicing of XBP1 was quantified using Image J software.

4.13 bZIP60 splicing analysis

Nicotiana benthamiana leaf discs expressing GFP or GFP-CaeB were floated on 10 μg/ml TM solution or as control on 0.1% DMSO. After 0, 2, 4 and 6 hr treatment total RNA was isolated using Z6-buffer (8 M guanidinium chloride, 20 mM MES, 20 mM EDTA, 50 mM 2-mercaptoethanol, pH 7.0). cDNA synthesis was performed using RevertAid Transcriptase (Thermo Fischer Scientific). bZIP60s was analysed with specific primers (5′-GGGGTTAGTTCTCCAGTGTTGTC and 3′-AGGGAACCCAACAGCAGACT) and either analysed via RT-PCR or via q-RT-PCR using the AriaMx Realtime PCR system (Agilent). Actin was used as control (5′-GGCTGGATTTGCTGGTGATG and 3′-TCCTTCTGTCCCATTCCGAC). Relative bZIP60s gene expression was analysed using the delta delta ct method as described by (Livak & Schmittgen, 2001).

4.14 Immunoblotting

For immunoblotting analysis cells were lysed in 2× SDS sample buffer, proteins were separated using 4–12% Bis-Tris Gels (Thermo Fischer Scientific) and transferred to a nitrocellulose membrane (GE Healthcare). Membranes were probed with the following antibodies: anti-cleaved PARP (cell signalling 5,625, 1:1,000), anti-XBP1s (cell signalling 12,782, 1:1000), anti-actin (Sigma-Aldrich A2066, 1:5,000), anti-Phospho-p38 MAPK (cell signalling 4,511, 1:1,000), anti-p38 MAPK (cell signalling 8,690, 1:1,000), anti-GFP (Thermo Fisher A6455, 1:1,000), anti-ATF6 (BioLegend 853,101, 1:500), anti-ATF4 (cell signalling 11,815, 1:1,000), anti-Flag M2 (MERCK F3165, 1:3,000) and anti-GAPDH (cell signalling 5,174, 1:5,000). Proteins were visualised by appropriate HRP-conjugated secondary antibodies. Densitometry analysis was performed using Image J software.

4.15 Immunofluorescence of infected MEF's and HeLa229

MEF's were fixed with 4% paraformaldehyde and permeabilised with 0.1% Trition X-100 before incubation with antibodies against Flag (staining tagged CaeB, labelled with Alexa Fluor 594) and Calnexin (staining the ER, labelled with Alexa Fluor 488). HeLa229 were transfected 18 hr prior immunostaining with GFP-CaeB using Lipofectamine 2000 (Thermo Fisher 11668019) according to the manufactures protocol. For mitochondrial staining the cells were incubated the last 45 min with Mitotracker CMXRos at 1:5,000 dilution of the stock according to the manufacture (Thermo Fisher). HeLa229 were fixed with 4% paraformaldehyde, permeabilised with 0.1% Triton X-100, and incubated with an antibody against Calnexin (staining the ER, labelled with Alexa Fluor 594). The cells were mounted using ProLong Diamond with DAPI to visualise cell nuclei and bacterial DNA using a Carl Zeiss LSM 700 Laser Scan Confocal Microscope and the ZEN2012 (black edition) software.

4.16 Infection of G. mellonella

Galleria mellonella larvae were purchased from TruLarv (UK). Larvae were infected by injection of 20 μl PBS, C. burnetii or C. burnetii ΔcaeB solutions (5 × 10⁴/μl) into the uppermost right proleg. The larvae were incubated at 37°C for 10 days and survival was monitored every 24 hr.

4.17 Immunofluorescence of infected hemocytes

At day 1 and 5 post-infection hemocytes from 3 G. mellonella larvae infected with C. burnetii or C. burnetii ΔcaeB were seeded and centrifuged on Poly-L-lysine coated coverslips. The coverslips were incubated with antibodies directed against C. burnetii (labelled with Alexa Fluor 488) and actin (phalloidin labelled with Alexa Fluor 647). The cells were mounted using ProLong Diamond with DAPI to visualise cell nuclei and bacterial DNA. The percentage of infected cells was determined by analysing at least 100 cells each in three independent experiments using a Carl Zeiss LSM 700 Laser Scan Confocal Microscope and the ZEN2012 (black edition) software. For the area measurement of the CCV's combined tile scan and Z-stack pictures were taken by the LSM 700 and 100 CCV's each in four independent experiments were analysed. To measure the XY dimensions, the longest distance and the corresponding 90° angle were determined with the profile module of the confocal software Zen2012 (Zeiss). The XY dimensions of each CCV were multiplied and plotted as Vacuole size with GraphPad Prism 5 (GraphPad software, San Diego).

4.18 Quantification of bacterial load/larvae

Day 1 and 5 post infection two infected G. mellonella larvae were disrupted using a BeatRuptor and lysed with Proteinase K overnight. Genomic DNA of C. burnetii was isolated with the Qiagen DNeasy Blood and Tissue Kit. RT-PCR was performed to determine bacterial load as described (Schulze-Luehrmann et al., 2016). To calculate fold replication the genome equivalent of day 5 was compared to day 1.